AI is leaving a select few with millions. Techies are debating whether the rest will be relegated to a "permanent underclass." Patrick Pleul/picture alliance via Getty Images AI is rapidly building wealth for some techies, but others think that a "permanent underclass" is coming. VC Deedy Das sparked debate with an X post about the wealth divide, with critics arguing "underclass" fears are overblown. "The anxiety everyone feels is real, both in the Bay or outside of it," Das said in a text to BI, and the lesson is to "be patient." Act fast, or risk being relegated to the "permanent underclass." That's the latest worry reverberating around the techosphere. It's no secret that AI is creating massive amounts of wealth for some. The startup space is flush with cash, the AI labs are doling out large salaries to retain top talent, and a series of upcoming IPOs could turn average engineers into millionaires many times over. For those early to the AI gold rush, that's joyous. For others, it's a sign of growing class disparity. Menlo Ventures partner Deedy Das described the feeling in his viral X post. Roughly 10,000 people who worked for companies like OpenAI, Anthropic, or Nvidia can now retire, he wrote. The rest making under $500,000 can work "their whole life and never get there." Das referenced the talk of a "permanent underclass" that flies around tech circles. It has long been a meme (and a warning) on X, as techies debate how artificial intelligence will reshape the economy and whether time is running out to set oneself up financially before AI outpaces human abilities. But Das' post clearly struck a nerve. Whether a "permanent underclass" is approaching may depend on who you ask. Are fears of a 'permanent underclass' overblown? Founder Tracy Chou compiled a list of words she felt are overused in San Francisco. Among "agency" and "taste," there was the phrase "permanent underclass." The backlash to Das' post was almost immediate. Replit product lead Amadeo Pellicce created a list of things to do to "thrive in SF as a member of the permanent underclass." They included going to therapy, living with others, and spending time with family. Then, he included a reminder: "X is not real life." "There is no permanent underclass," Pellicce wrote. "Labor can always create capital. All you have to do is make something people want. AI just accelerates this." The permanent underclass of SF can go cry about not having pre-IPO anthropic money on their bi-annual trip to Europe or Japan this summer — cato 😾 (@chowtato) May 17, 2026 Foundation Capital partner Jaya Gupta wrote that we may not want everything the so-called overclass receives. "They cannot monopolize joy, taste, friendship, beauty, aliveness, or the reasons life is worth living," she wrote. In a text to Business Insider, Das acknowledged that some people had reacted with some version of "quit your whining" or "touch grass." Das' personal reaction: "The anxiety everyone feels is real, both in the Bay or outside of it." "This is just a side-effect of capitalism," Das wrote. "Any new technology has ever increasingly faster adoption rates and thus increasingly faster rates of wealth building." Das' post was specifically about San Francisco. Former Facebook product manager and "Chaos Monkeys" author Antonio García Martínez wrote that it correctly captured the city's "ruthless boom/bust, winner-take-all competition." Some Bay Area-lovers pushed back. Notion marketer John Hurley fought back with a list of beautiful things you can do in the city. "Be deeply grateful for the heavenscape you live in," he wrote. "Permanent underclass?" tech PR Ed Zitron wrote on X. "Drive out to the Redwoods, breathe in the air." did you escape the permanent underclass yet pic.twitter.com/2wLlvauUdW — taoki (@justalexoki) May 11, 2026 Meanwhile, techies have started diagnosing the traits of future members of the underclass, such as using a notetaking app or saying "I think." Real career concerns beneath the memes While it might be easy to dismiss the warnings of a "permanent underclass," the memes are also hitting on real fears people have for their careers. AI leaders keep forecasting how radically the technology will shake up the economy. It could be a white-collar bloodbath, as Anthropic CEO Dario Amodei puts it. Meanwhile, layoffs keep getting blamed on AI. Others talk about the idea that AI will eventually be so capable that we'll all just be collecting universal basic income. Americans' sentiment for AI data center development has nosedived. One user posted a clip of ex-Google CEO Eric Schmidt's commencement speech, where graduates booed his reference to AI. "These kids are embracing the permanent underclass," they wrote. It may also be localized in San Francisco (and its famously high cost of living) — and San Francisco alone. One X user wrote that the permanent underclass was real, but that non-tech people should ignore it. "What SF people consider 'permanent underclass' means something very different than in the rest of the country," they wrote. "But it means 'relative permanent underclass,' and it is real." In his text, Das wrote that many San Francisco residents had been supportive, and that the common reaction he'd heard from those in the city was: "Thank you for stating plainly the anxiety from conversation we've all been having behind closed doors but too afraid to say out loud." you are in the permanent underclass if you can't afford the AI subscriptions — shafu (@shafu0x) May 10, 2026 Another user wrote that AI was both minting checks and threatening the "safe path" — something they said didn't happen during previous technology booms. "You could sit out the dot-com boom & keep your accounting job," they wrote. Kierra Dotson, author of The Data Digest, suggested there is a middle ground to the discourse. "I think there's a lot of room between believing you're free of money from money problems and about to be in the permanent underclass(severe). Homes being expensive in the Bay Area is nothing new(esp during the COVID peak) and the median salary is $136k," she wrote. "Has the goal post moved? Absolutely. Are these people in golden handcuffs? Probably," she added. "But I think they have many more options than they think they do and expectations will have to be reworked…which is something we are all waking up to." Das wrote in his text that the lesson was to "be patient." "If you look at the people who were the biggest beneficiaries of the AI era, it is people who were diligently doing research since 2013/4, when their skills weren't 'hot,'" he wrote. "Patience is underrated." Read the original article on Business Insider

Venture capitalist Chamath Palihapitiya said it's a "failure" if consulting firms don't recognize what OpenAI and Anthropic are trying to do. Jordan Strauss/Invision/AP Venture capitalist Chamath Palihapitiya has a warning for consulting firms. Palihapitiya said companies like PwC and Accenture should be cautious about working with OpenAI and Anthropic. He said such deals are like "letting the fox into the hen house." Venture capitalist Chamath Palihapitiya said leading consulting firms will come to regret their partnerships with OpenAI and Anthropic. "If you are running a consulting business and you are deploying Anthropic or OpenAI directly into your organization (I'm looking at you PwC and Accenture) you are letting the fox into the hen house," Palihapitiya wrote on X. If you are running a consulting business and you are deploying Anthropic or OpenAI directly into your organization (I’m looking at you PwC and Accenture) you are letting the fox into the hen house. OpenAI and Anthropic are openly funding and starting competitors to you while… https://t.co/MYWEeSjnVD — Chamath Palihapitiya (@chamath) May 17, 2026 Palihapitiya issued his warning days after OpenAI announced that it was forming a new company, OpenAI Deployment Company, "to help organizations build and deploy AI systems." McKinsey & Company is an investor in DeployCo. "OpenAI and Anthropic are openly funding and starting competitors to you while also using your usage to drive more success for them," he wrote. "This is not a failure on their part but a failure on your part. " PwC and Accenture have previously announced expansive partnerships with OpenAI and Anthropic. Last week, Anthropic and PwC jointly announced "an expansion of their strategic alliance," including the establishment of "a joint Center of Excellence and a program to train and certify 30,000 PwC professionals on Claude." On the same day, May 14, Accenture announced a partnership with OpenAI to help federal agencies deploy AI calls for the type of forward-deployed engineers that Palihapitiya seems particularly concerned about. All of this, Palihapitiya said during a recent episode of the "All in Podcast," is about the realization that the low end of the SaaS market is "basically finished," thanks to generative AI. Now, the competition is about the big companies. "At the high end of the market where all the action is, what people are finding is, 'Hey, hold on a second, this is a lot harder than we thought,'" Palihapitiya said. "It's not like boop, boop, boop, put in a prompt and beep, bap, boop, it all works. It's not how it works." It's why AI companies need organizations like DeployCo, Palihapitiya said, and why consulting companies should be wary of short-term partnerships. Otherwise, as he wrote on X, they are just "adopting and accelerating the companies that want to disrupt them." Read the original article on Business Insider

Joyce Hankins / Unsplash The pre-print website arXiv has announced that researchers who put their names to papers which included errors clearly generated by artificial intelligence (AI) will face a year-long ban and ongoing restrictions. The move is a response to a growing influx of AI-generated papers faced by scholarly journals as well as sites such as arXiv, which serve as unofficial platforms for research publication ahead of peer review. However, not everyone agrees that arXiv’s response to the problem is appropriate – and the solution to the flood of AI slop research may involve more AI, not less. The rise of bot-assisted writing AI-generated text is on the rise everywhere. A study released last week suggests half of new articles published online are now “primarily AI-generated”. Science is not immune to this trend. Last month, the journal Organization Science published a study of how the rise of AI has affected submissions and peer reviews since the release of ChatGPT in 2022. Reporting a dramatic rise in submitted papers and a drop in quality, the authors conclude that “the current state of AI tools, amplified by existing publish-or-perish incentives, appears to be pushing the system toward an equilibrium of more rather than better research”. A common problem in AI-generated research writing is hallucinated citations: references to other research that does not exist. The traditional safeguard against poor quality in scholarly publishing is peer review: another expert in the subject at hand reads the research paper and interrogates the work behind it before it can be published. However, the peer review system was already struggling before AI. Pressured researchers often have little time or incentive to do the unpaid work of peer review. And on arXiv, which publishes preprints – articles which have most often not been peer-reviewed – even this system is not available. Last year, flooded with AI-generated submissions, the site stopped accepting certain types of article. A study published in January (itself a preprint) estimated around 1 in 8 papers in biomedical science now contain AI-generated text. Most researchers would agree that AI-generated text is not a problem in itself. The problem is the lower-quality work that AI can make easy to produce. Does the punishment fit the crime? The ArXiv announcement doesn’t come out against AI use, but rather says If a submission contains incontrovertible evidence that the authors did not check the results of LLM generation, this means we can’t trust anything in the paper. This may be true as far as it goes. But the penalty – a year-long ban for all authors listed on a paper – may be out of keeping with current research practices. In the past, research was often carried out by people working alone or in groups of two or three. In these circumstances, it seems reasonable to expect each author to take responsibility for the whole. But research is now more collaborative than ever before. Many papers have four or five authors, and in a growing number of extreme cases papers may be credited to groups of hundreds of scientists working together, each working on their own speciality and trusting their colleagues to be doing the same. In a case where one author of dozens or hundreds included an AI-hallucinated reference in their part of the paper, banning the lot seems harsh. And there are no equivalent sanctions for publishing other problematic material. There’s no ban for pushing fringe or discredited theories, or using poor quality evidence and illogical arguments, for example. Can AI help fight slop? The rise of AI produces problems for publishers and quality assurance. And the idea of some kind of sanctions for reckless use of AI, such as included hallucinated references, is a good one. But ArXiv’s particular choice seems drastic. If the goal is to improve peer review and quality assurance, AI systems themselves can play a role. Modern AI systems are quite capable of taking a list of references and checking everything on it is a real paper available on the internet. Any references flagged as suspect can then be checked by a human. AI can even be useful for carrying out quick sense-checks of things like a paper’s statistical analysis. Perhaps this is the way forward, rather than harsh sanctions for relatively minor AI-related infractions. Vitomir Kovanovic does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

Venti Views/Unsplah, FAL The Academy of Motion Picture Arts and Sciences has adjusted the eligibility criteria for films vying for Oscars from 2027 onward. Films featuring actors generated by artificial intelligence (AI) are now ineligible, as are scripts that aren’t demonstrably human-authored. Crucially, the rules do not ban AI – generative or otherwise – altogether. The Academy explicitly acknowledged the widespread adoption of generative AI, and has left it to voters to determine whether a film’s creative direction is substantively driven by humans. Academy president Lynette Howell Taylor framed it simply: “humans have to be at the centre of the creative process”. The rules were imposed following specific controversies: the 2025 awards season surfaced AI voice modification in The Brutalist, AI voice cloning in Emilia Pérez, and varying degrees of AI use in A Complete Unknown and Dune: Part Two. The resulting public debate has focused almost entirely on what audiences can see: generated faces, synthetic voices and digital resurrections. But this focus ignores the main areas of film production in which AI actually plays a key role. Cinema’s automation history Automation tools have been embedded in cinema for longer than most people realise. When non-linear editing software Avid Media Composer launched in 1989, it replaced the physical cut-and-splice flatbed editing process that had defined post-production for decades. This all but eliminated traditional linear editing suites, and the skilled labour that went with them, within a few years. Computer-generated imagery (CGI) entered mainstream production with Tron (1982) and Jurassic Park (1993). And by the mid-2010s, deepfake and neural rendering technology (distinct from generative AI) made it possible to manipulate human likenesses directly – such as de-aging Jeff Bridges in the Tron sequel, Tron: Legacy (2010), and Michael Douglas and Kurt Russell in the Marvel Cinematic Universe. Generative AI aids and improves these existing processes. Director Robert Zemeckis used AI company Metaphysic to de-age Tom Hanks and Robin Wright in real time, on camera while shooting Here (2024). And the forthcoming film As Deep as the Grave will feature an AI-generated likeness of Val Kilmer, who died in 2025. This is the most prominent case yet of AI being used to “generate” a performer who couldn’t physically be present. These examples sit on an ethical spectrum, from de-aging a living actor to synthesising a performance based on a deceased star. The Academy’s new rules attempt to draw a line somewhere on that spectrum. Where exactly that line should go is an ongoing question. Where AI is usually used in film The most pervasive uses of AI in contemporary mainstream productions are usually invisible to audiences. For most studio-produced feature films, AI-assisted platforms handle script breakdown in pre-production. They help with extracting production requirements based on the screenplay, scheduling shoots and modelling budgets. These tasks once took hours, or even days, to complete. On set, AI may be used to generate the background environments for virtual production stages. Originally used for TV shows such as The Mandalorian, these sets are increasingly used as a cost-effective option in studio productions. In post-production, AI tools handle first-pass editing, audio clean-up, VFX, and voice modification. For instance, Dune: Part Two (2024) used machine learning to achieve the blue-eye effect – a minor enhancement audiences don’t immediately register. Arguably the most widespread (yet invisible) use of AI is for localisation, as streaming platforms pursue global audiences at scale. AI-assisted dubbing and subtitle generation are now standard processes. AI can also influence what films get made. Companies use algorithmic analysis to inform greenlight decisions, predict trailer performance, and optimise release windows. As Vue Cinemas CEO Tim Richards put it, “[AI] determines what we show at what cinema, on what screen and at what time”. Automation and AI-driven processes have a long history in film. Yet it is only now, amidst the deluge of AI-generated content, that these modes of filmmaking have attracted the public’s attention. Not a new problem, but a newly visible one The Academy’s changes do not reflect a resistance to innovation – but a centring of human authorship. The most publicly legible creative contributions to a film should remain demonstrably human in origin. That’s a reasonable position, with broad support across audiences and industry professionals alike. But it sits alongside the harder question of consent: not just whether AI was used, but whether the people whose likenesses, voices, and labour are implicated actually agreed to it. The Oscars’ rules are about disclosure as much as eligibility – and they aren’t alone. Many publishers and creative bodies are now formulating AI principles that emphasise the need for transparency. The question isn’t whether AI is being used in film and TV production; it is. The question is whether audiences have the right to know — and whether public discourse has caught up enough to know what it’s even asking about. Generative AI is the latest inflection in a long history of automation — more powerful and fast-moving than its predecessors, but no different in the questions it raises. The pressing task now is making sure the frameworks for consent, disclosure and creative attribution keep pace with AI technology. Daniel Binns is an Associate Investigator with the ARC Centre of Excellence for Automated Decision-Making and Society. Meg Thomas is a student member of the ARC Centre of Excellence for Automated Decision Making and Society.

Explore seven AI photo trends powered by Google Gemini, from cinematic edits to intentional imperfections, and how prompts are reshaping creative expression in 2026. The post 7 Best Gemini Photo Editing Trends in 2026 (With Prompts to Try) appeared first on TechRepublic.

Kirsten Korosec / TechCrunch: Anthropic acquires NYC-based Stainless, which generates SDKs from APIs, and plans to wind down its hosted products, after reportedly discussing a $300M+ deal  —  Anthropic announced Monday it has acquired Stainless, a startup founded by former Stripe engineer Alex Rattray whose software …

The release of Shai-Hulud source code spells trouble for software developers as researchers worry the self-replicating worm could scale.

"I think there's plenty of fire lit under them already."

Stainless, a New York-based startup, founded in 2022, rose to prominence in the emerging AI industry for automating the creation and maintenance of software development kits, or SDKs — the libraries developers use to interact with APIs.

Jackson Chen / Engadget: X quietly limits users who didn't pay for verification to “50 original posts and 200 replies per day”, down from 2,400 posts per day  —  That's down from the previous 2,400 posts per day limit.  —  X has introduced some more incentive to get users to pay for “verification” …

Natalie Lung / Bloomberg: Uber increases its stake in Delivery Hero to 19.5%, up from 7% in April, and says it “has no intent to acquire 30% or more” of Delivery Hero's voting rights  —  Uber Technologies Inc. has increased its stake in German food delivery company Delivery Hero SE, underscoring its ambitions …

Kevin "Mr. Wonderful" O'Leary is encountering vocal opposition to his plans for a huge data center in Utah. VALERIE MACON / AFP via Getty Images Americans don't agree on much. But a majority of them don't want to live near a data center. That's a real problem for AI boosters. Analyst Ben Thompson has a suggestion: Cut everyone — everyone — a check. Kevin O'Leary makes a great villain. The celebrity investor spent decades honing his image as the least likable rich guy on "Shark Tank," reveling in what he calls brutal honesty. Last year riffed on that persona in "Marty Supreme," highlighted by a scene where he spanks Timothée Chalamet with a ping pong paddle. So it's not surprising that he's just emerged as the hateable face of the AI Data Center Backlash, as he promotes a massive development in Utah. Stories about opposition to the O'Leary-backed, 40,000-acre Stratos Project have hit national outlets and social media in the last month. Tucker Carlson, who knows a hot button when he sees one, had O'Leary on his show last week, setting him up as a real-life Mr. Monopoly out to exploit Utah taxpayers. Now O'Leary is arguing that people who don't like his project are professional protesters, funded via shadowy boogeymen. But the details about O'Leary's project and the people who hate it are beside the point: Opposition to data centers is a widespread, bipartisan phenomenon throughout the US, as recent Gallup polling shows. The technocratic backlash to the backlash is to tell people who don't like data centers that they're wrong, and that data centers don't really hog precious water or energy. (Business Insider published a prize-winning series about these claims and counter-claims last year.) But this one feels like it's broken containment. Data centers are convenient repositories for everyone's fears and anxieties about AI — not just what it may or may not do to our environment, but to what it will do to everything, starting with our economic future. Again: That anxiety is totally reasonable. Because everyone — starting with the people running the biggest AI companies — predicts that AI will unleash massive changes in the workforce. We're also told that all of this is inevitable, and that we don't really have a say in it — you can't hold back technology! — and that all we can do is hope to adapt. So pushing back against a data center project in your town seems like a pretty reasonable kind of protest vote. Maybe I can't keep AI from upending my life, but at least I can keep it out of my backyard. So what if you're a well-meaning AI booster who thinks the backlash is truly misguided, and that fighting against data centers is as dumb as fighting against highways? Analyst Ben Thompson has a suggestion: Pay off the backlash by cutting checks to people in data center towns. If data centers are a resource for our AI future, then start paying people for that resource. If that data center up the road weren't sold to my neighbors based on amorphous tax benefits that my local government may or may not spend appropriately, but rather were to result in a check in the mailbox every year, I suspect you could get a lot more people on board! The sober answer to this, of course, is that AI data centers are already supposed to be paying the people in the towns they're in. They bring jobs in the short term, and are meant to spur all kinds of economic activity in the long term. But it's one thing to tell someone the ominous-looking building down the street is going to make some electricians a bunch of money. It's another to hand everyone a check. And if it's the Kevin O'Learys and other technocrats writing the check — taking cash from their pockets and putting it in yours — maybe it makes the worrisome future of AI easier to handle, in the present tense. Read the original article on Business Insider

For some ancient Aboriginal Australian communities, dingoes were part of the family.

Discussion | Link

CJ Haddad / CNBC: Seagate shares drop 9%+, leading a group-wide sell-off after comments from its CEO raised concerns it won't be able to meet demand fueled by the AI buildout  —  Shares of memory chip maker Seagate slipped more than 8% Monday, leading a group-wide sell-off after comments from CEO Dave Mosley raised concerns …

Biologists use Co-Scientist to find novel factors that successfully rejuvenate human cells.

Only a couple vendors could likely fulfill what the FBI is after, namely Flock and Motorola.

Musk plans to appeal after judge immediately affirmed the jury's decision.

Cristian Dina / The Next Web: Sigma, which sells a cloud-native analytics platform that sits on top of data warehouses, raised an $80M Series E led by Princeville Capital at a $3B valuation  —  The round was led by Princeville Capital, with new strategic investors Databricks Ventures, ServiceNow Ventures, and Workday Ventures also participating.

New X-rays and CT scans showed small case with locking mechanism containing metal instruments.